2. object appears to travel in the same direction as
the tube shift and the object closer to the tube
appears to move in the opposite direction [the so
called Same Lingual Opposite Buccal (SLOB)
rule; this could equally be remembered as Buccal
Opposite Palatal Same (BOPS)]. The parallax
technique may also be applied when the radio-
graphs are taken at a different vertical angulations
(vertical parallax). Various combinations have
been used based on the technique of parallax:
(1) two intra-oral periapical radiographs taken
at different horizontal angles (Clark, 1909).
(2) one maxillary anterior occlusal and one
maxillary lateral occlusal (Southall and
Gravely, 1989).
(3) one periapical and one maxillary anterior
occlusal radiograph (vertical parallax; Rayne,
1969).
(4) one panoramic and one maxillary anterior
occlusal radiograph (vertical parallax; Keur,
1986).
Although the tube for the panoramic radiograph
is actually positioned behind the patient’s head
with a vertical angle of –7 degrees, in order to
explain the shift technique it may be considered
to be in front of the head at an angle of +7
degrees. Since the maxillary occlusal is taken at a
vertical angle in the range of 60–65 degrees,
there is an effective difference ranging from
53–58 degrees between the two films as shown
in Figure 1 (Jacobs, 1994).
(5) one panoramic radiograph alone when a
Panorex machine is used (Turk and
Katzenell, 1970).
Magnification
This technique is based on the principle of ‘image
size distortion’; that is for a given ‘focal spot’–
film distance, objects further away from the image
receptor (film) will be depicted more magnified
than objects closer to the film (Figure 2; Goaz
and White, 1982). Two methods can be used with
this technique.
Status-x radiography (Ostrofsky, 1976). This
involves a special tube inserted in the patient’s
mouth and a film held around the face by the
patient. The hollow anode provides nearly a
point source of radiation and, therefore, the laws
of central projection apply—the object closer to
the source is magnified.
Panoramic radiography (Wolf and Matilla, 1979;
Fox et al., 1995). The principle of image dis-
tortion can be applied in panoramic radiography.
If a canine is relatively magnified in comparison
to the adjacent teeth in the arch or the contra-
lateral canine, it will be located closer to the tube,
i.e. palatally, and if it is relatively diminished it
will be located further away from the tube, i.e.
labially. This method is most effective when the
canine is not rotated, not in contact with the
incisor root and the latter is not tipped
Other methods reported in the radiographic
localization of the impacted canine are the vertex
occlusal radiograph (Hitchin, 1951), image
sharpness, and relationship of the canine cusp tip
with the lateral incisor root in the panoramic
radiograph. The ‘right angle’ technique involves
two films taken at right angles to each other so
that the canine can be located in three dimensions,
e.g. lateral skull and postero-anterior cephalo-
gram (Broadway and Gould, 1960), or lateral
skull and panoramic radiograph (Coupland,
1987). Tomography, e.g. polytomography (Ericson
and Kurol, 1987), and computer tomography
(Ericson and Kurol, 1988) are especially useful in
cases of root resorption of the adjacent teeth.
26 C. MASON ET AL.
Figure 1 Vertical angulation between the orthopantomo-
gram and the maxillary anterior occlusal.
3. Radiographic localization of the impacted
canine is very important, especially if surgical
treatment is to be undertaken. For many cases,
radiography is the only means of accurately
determining the position of the unerupted canine.
The fact that so many radiographic localization
techniques exist shows that none of them is ideal.
A study on the radiographic prescribing habits of
orthodontists and oral surgeons showed that
78 per cent used more than two radiographs and
23 per cent used four or more (Southall and
Gravely, 1987). It may be possible that clinicians
use the technique with which they are most
familiar at the cost of an increased dose of
radiation to the patient (Southall and Gravely,
1987). An increase in the level of dental radio-
graphy by a factor of six between 1957 and 1977
has been reported (Kendall et al., 1981).
Clinicians should be encouraged to evaluate
the radiographic methods currently in use in
terms of the dose levels to which the patients
are exposed. If it can be shown that a single
radiographic technique provides the same
information as other techniques that require a
higher radiation exposure, the justification for
the use of those with higher exposure must be
questioned (Southall and Gravely, 1987).
The panoramic radiograph is widely used in
general practice and is the first choice radiograph
for orthodontic patients (Southall and Gravely,
1987). Use of the maximum amount of
information that can be obtained from such
a radiograph would help reduce unnecessary
radiation exposure to patients.
The current study was undertaken in order to
assess the validity of two different radiographic
techniques in the localization of the impacted
canine: vertical parallax (from a panoramic
and anterior maxillary occlusal radiograph) and
magnification (from a single panoramic radio-
graph). The null hypothesis tested was that there
was no difference between the two techniques.
RADIOGRAPHIC LOCALIZATION OF CANINES 27
Figure 2 Increasing the distance between the object and the film increases the
magnification.
4. Materials and methods
The study consisted of a retrospective analysis of
the records of 100 patients who had undergone
surgical treatment of unerupted canines in the Day
Stay Unit of the Eastman Dental Institute, London.
The following criteria were used in order to
select the sample:
1. For every patient included there should be
one panoramic and one maxillary anterior
occlusal radiograph of acceptable diagnostic
quality (e.g. no distortions on the panoramic
film due to incorrect patient positioning).
2. If a patient had gross distortion of the dental
arches due to a cleft palate they were
excluded.
3. Ideally, the radiographs should have been taken
on the same day, shortly before the operation.
If the two radiographs were taken more than
6 months apart or active orthodontic treatment
had taken place between the two radiographic
examinations, the patient was excluded.
4. A surgical diagnosis of the position of the
impacted canine should be clearly stated in
the operation records otherwise the patient
was excluded.
Radiographic examinations
The estimated position of the canines with the two
different radiographic techniques was recorded
for six examiners: a consultant maxillofacial
surgeon (A), a consultant orthodontist (B), a
consultant in paediatric dentistry (C), a junior
member of staff (senior house officer) (D), and
two MSc postgraduate students (E and F). One
postgraduate (the person responsible for the
study) assessed the whole sample, whilst the rest
examined a sub-sample of 40 patients. The exam-
inations were standardized: the radiographs
were studied on the same conventional X-ray
illuminator (Kodak Coldlight Illuminator series 2,
Ostsiodern Scharnhausen, Germany). Black card
templates were used in order to eliminate the
peripheral light from the X-ray viewer and the
names, hospital numbers, and dates of birth on
the radiographs were hidden. All radiographs
were taken at the Department of Radiography,
Eastman Dental Hospital. The maxillary anterior
occlusal radiographs were taken using a Philips
X-ray machine (50–65 KV and 7.5 mA; Philips
Medical Systems, London UK) using an intra-oral
film (Kodak E-speed Plus). The panoramic radio-
graphs were taken by either Planmeca 2002CC-
Proline (66 KV and 8 mA; Planmeca, Helsinki,
Finland) or Siemens Orthopantomograph 5 (65–70
KV and 15 mA; Siemens AG, Munich, Germany)
using an extra-oral film AGFA HT-U (green
sensitive; AGFA-Gaevert, Antwerp, Belgium).
Each of the examiners examined the
radiographs on two separate occasions using
the two different techniques. The familiarity of
each examiner with the techniques was obtained
with a questionnaire, since this could be helpful
in explaining differences in accuracy between the
two techniques. At the first examination, the
canines were localized with vertical parallax
from the panoramic and the anterior occlusal
radiographs. An example is given in Figure 3a,b:
the impacted right maxillary canine has moved
upwards (‘with the tube’) from the orthopanto-
mogram to the anterior occlusal radiograph and
therefore is localized palatally. At the second
examination (1 month later), the localization
of the same canines was undertaken with the
magnification method by the panoramic radio-
graphs only. For example, in the previous ortho-
pantomogram (Figure 3a), the impacted canine is
horizontally magnified with comparison to the
contralateral erupted canine and is therefore
localized palatally. The following guidelines
were given prior to the examination if the
examiners were unfamiliar with the magnifica-
tion technique (Fox et al., 1995).
The impacted canine will be classified as:
(A) Buccal: Where relative diminution of the
crown has occurred in the hori-
zontal plane, compared with
adjacent teeth lying in the arch.
(B) In the arch: Where the canine has the same
degree of horizontal magnifica-
tion as adjacent erupted teeth.
(C) Palatal: Where relative magnification
of the crown has occurred in the
horizontal plane compared with
adjacent teeth lying in the arch.
28 C. MASON ET AL.
5. In case of unilateral impaction the comparisons
should be carried out with the contralateral
normally occluded canine.
Each examination was blind, i.e. the examiners
did not know the surgical outcome.
Reproducibility studies
One week after each examination a randomly
pre-selected sub-sample was re-examined in order
to assess intra-examiner variation. This sample
was selected from random number tables, and
consisted of 40 patients for the whole sample and
16 patients for the smaller sample of 40 patients
(40 per cent).
Statistical analysis
The predicted positions of the impacted canines
from each radiographic technique were compared
with the actual positions found during surgery.
The overall percentages of correct diagnosis for
each technique were compared with a SND test
of statistical significance. The purpose was to
determine the extent to which the two methods
differed. A Cohen’s Kappa statistic was also
used in order to compare agreement of each
technique with the true position of the canine.
Validity of diagnosis (i.e. correct diagnosis) with
either the magnification or parallax technique
was expressed as sensitivity for the palatally and
RADIOGRAPHIC LOCALIZATION OF CANINES 29
Figure 3 (a) Orthopantomogram showing an impacted right maxillary canine. (b) Maxillary anterior
occlusal radiograph of the same patient as 3(a).
6. buccally located canines. Sensitivity for the
palatally located canine is the probability that a
palatally placed canine will be correctly detected
with the specific technique.
The above statistical analysis was carried out
considering each patient as a single unit and
only one canine from each patient was analysed.
Therefore, in the case of bilaterally impacted
canines the right or the left canine was randomly
chosen for analysis. This was undertaken in
order to eliminate possible bias from a statistical
analysis of variables that were not completely
independent, i.e. two impacted canines that
belong to the same patient.
Intra-examiner reproducibility for each
technique was also tested with a Cohen’s Kappa
statistic.
Results
The records of 100 patients (71 females, 29
males) with a total of 133 impacted maxillary
canines were used for the study. Thirty-three
patients had bilaterally impacted canines. Of
the 133 canines, 87 (65 per cent) were located
palatally, 38 (28 per cent) buccally, and eight
(6 per cent) in the line of the arch.
Comparisons between true position and
predicted position from the radiographic
examinations
The results from the radiographic examinations
with the two different techniques were compared
with the surgical results by a Cohen’s Kappa
statistic. The Kappa statistic, between true canine
position and predicted position with parallax,
had a range of 0.13 to 0.46 for the six different
examiners. Similarly, the Kappa statistic between
true position and predicted position with
magnification had a lower range of 0.06 to 0.26
for the same examiners (Figure 4).
With regard to the surgery, the ‘buccal’ and
‘in the line of the arch’ positions required a
buccal flap (for surgical exposure or extraction
of the impacted canine). The two positions were
therefore combined and new Kappa values were
calculated. This resulted in higher Kappa values
for most of the examiners for both parallax and
magnification (Table 1).
Comparisons between parallax
and magnification
From the previously calculated Kappa values,
a difference between the parallax and the
magnification method was evident. Comparing
the percentages of agreement with the two
methods for examiner F, derived from Tables 2
30 C. MASON ET AL.
Figure 4 Kappa statistic between true position and
position predicted by parallax and magnification for the six
examiners.
Table 1 Kappa values before and after ‘buccal’ and ‘in the arch’ positions were combined for parallax and
magnification (all examiners).
Examiner code Parallax Parallax Magnification Magnification
Normal Kappa Kappa when Normal Kappa Kappa when
buccal = arch buccal = arch
A 0.29 0.41 0.10 0.24
B 0.17 0.48 0.17 0.30
C 0.46 0.51 0.26 0.31
D 0.25 0.32 0.06 0.07
E 0.13 0.33 0.13 0.08
F 0.45 0.50 0.20 0.29
7. and 3, the SND statistical test had a P value
0.05 < P < 0.1 and the confidence interval of the
true difference between the two methods ranged
from 0.016 to 20 per cent.
Sensitivity for the palatally or buccally located
canine
The sensitivity of both parallax and mag-
nification in the detection of the palatally or
buccally located canine for examiner F, is shown
in Figure 5. Both methods were less sensitive
in detecting the buccal canine, especially the
magnification.
Reproducibility studies
Intra-examiner reproducibility was assessed
with a Kappa statistic for both parallax and
magnification, for each examiner (Table 4).
Table 5 shows the familiarity of each examiner
with the parallax technique in comparison with
their Kappa scores. This could not be carried out
for magnification since none of the examiners
was familiar with the technique.
Discussion
The sample for the study consisted of patients
who had undergone surgery under general
anaesthesia for an impacted canine and for whom
a panoramic and an upper anterior occlusal
radiograph had been taken. Most of the patients
had been referred by the Department of
Orthodontics at the hospital where the two
radiographs are routinely taken for each patient
with an impacted canine.
The correct diagnosis of the impacted canine
position with the vertical parallax varied quite
markedly between the six examiners. The best
Kappa score was achieved by the consultant in
paediatric dentistry (0.46) and the worst by one
MSc student (0.13). Even for the highest scores
RADIOGRAPHIC LOCALIZATION OF CANINES 31
Table 2 A comparison between true canine position
and position predicted by parallax for examiner F.
Position predicted by parallax
Arch Palatal Buccal Total
True Arch 2 2 1 5
Position Palatal 3 62 4 69
Buccal 3 11 12 26
Total 8 75 17 100
Table 3 A comparison between true canine
position and position predicted by magnification for
examiner F.
Position predicted by magnification
Arch Palatal Buccal Total
True Arch 1 4 0 5
Position Palatal 6 62 2 70
Buccal 7 15 3 25
Total 14 81 5 100
Figure 5 Sensitivity for the palatal and buccal canine for
both parallax and magnification for examiner F.
Table 4 Kappa values from the parallax and
magnification reproducibility studies, for each
examiner.
Examiner code Kappa value Kappa value with
with parallax magnification
A 0.35 0.81
B 0.67 0.52
C 0.55 0.91
D 0.30 0.82
E 0.49 0.33
F 0.63 0.76
8. the agreement was only ‘moderate’ and, therefore,
not entirely satisfactory.
These results are not in agreement with a
study comparing vertical parallax with three
other radiographic techniques (Southall and
Gravely, 1989). In that study, although the Kappa
statistic was not calculated, the percentage of
agreement with the true position was 87 per cent.
In the current study, the agreement was 76 per
cent for examiner F. It should be noted that
comparisons between the two studies are not
appropriate, since one was a laboratory study
where a series of radiographs of a dried skull
with a radio-opaque marker placed in different
positions, were tested.
The large variation in the Kappa scores
between the six examiners could not be
explained from the familiarity with the vertical
parallax technique. The examiners with the best
scores only used the technique ‘sometimes’, and
the examiner who used the technique ‘routinely’
had the second worse score. This examiner was the
consultant orthodontist who, although diagnosing
the impacted canine, was not performing the
surgery. Nevertheless, when the ‘buccal’ and
‘arch’ positions were combined, the same exam-
iner scored very high. This indicates that he was
diagnosing many of the buccally positioned
canines to be ‘in the line of the arch’ and vice
versa. The combination of ‘buccal’ and ‘arch’
positions improved the Kappa scores of all
examiners, but the agreement was still only
‘moderate’.
It should be noted from the above that no
conclusions can be extrapolated for the different
dental specialties, since the differences may be
due to individual variation, and not necessarily
to differences in training and experience.
The correct diagnosis with the magnification
method also varied between the six examiners
and was less accurate than the diagnosis with
vertical parallax for most of the examiners. The
examiner with the best Kappa score was again
the consultant in paediatric dentistry (0.31), and
the worst score was the senior house officer (0.06).
For two examiners (the consultant maxillofacial
surgeon and one MSc student), the Kappa scores
were similar for parallax and magnification.
Even for the highest scores the agreement
was only ‘fair’ and therefore not satisfactory.
The combination of the ‘buccal’ and ‘arch’
position resulted in higher Kappa scores for the
magnification for almost all examiners, but the
agreement was still only ‘fair’.
In comparison with two previous investigations,
the magnification method was less successful. In
an earlier study (Wolf and Matilla, 1979) the
percentage of agreement was 88 per cent for the
first and 89 per cent for the second examiner.
In the present study, the agreement with the
magnification method for examiner F was only
66 per cent. However, the ‘gold standard’ for
the previous study was another radiographic
technique (similar to the horizontal parallax)
and not the true position of the canine, which, in
this study, was confirmed at operation. Variations
in the sample could also explain differences in
agreement; the above study had comparatively
more palatally impacted canines for which the
magnification has a high sensitivity (89 per cent
in the current study). In a later study (Fox et al.,
1995) the percentage of agreement was 76 per
cent with a Kappa of 0.52, while in the current
investigation the results were 66 per cent and
0.2, respectively, for examiner F. Again, the ‘gold
standard’ in the later study was a ‘vertex occlusal’,
which ‘have been shown to be unreliable under
certain circumstances’ (Southall and Gravely,
1987). In addition, in both of the previous studies
each impacted canine was considered as a unit
with the possibility of introducing bias.
Although the Kappa statistic shows that the
magnification method is less accurate than
vertical parallax, the difference in agreement
did not reach a significant level of P < 0.05.
32 C. MASON ET AL.
Table 5 Relationship between clinical experience
and Kappa values for parallax for each examiner.
Examiner code Clinical Kappa
experience value
A Never 0.29
B Always 0.17
C Sometimes 0.46
D Sometimes 0.25
E Often 0.13
F Never 0.45
9. Therefore, to subject a patient to higher radiation
exposure (with the inclusion of the anterior
occlusal radiograph) may not be justified in the
case of a palatally impacted canine, for which the
sensitivity is the same with both techniques (89
per cent).
In the study by Fox et al. (1995), the sensitivity
in the detection of the palatal canine with
magnification was also high (82 per cent), while
that for the buccal canine was lower (65 per
cent). In the current study, the sensitivity for the
buccal canine was very low (11 per cent).
Reproducibility studies
For the majority of the examiners, intra-examiner
consistency was greater with magnification
than with parallax. This was not expected since
magnification is considered to be more subject-
ive than parallax, but the fact that the method is
easier and simpler may have contributed to a
high reproducibility. The senior house officer
achieved the highest Kappa score for magnification
(0.82) and the consultant orthodontist the
highest score for parallax (0.67). In the study
by Fox et al. (1995), the reproducibility score
for magnification was similarly high (0.86).
The reproducibility scores for parallax are not
satisfactory overall and, although the method is
more successful in predicting the true position of
the canine, it is far from ideal.
The results are not directly related to the
diagnostic ability of the examiners; the overall
diagnosis of an impacted canine is also based on
clinical observations (i.e. intra-oral observation
and palpation). It is encouraging that the buccally
impacted canine, which is less successfully
detected with the two methods, can often be
palpated in the labial sulcus.
Conclusions
The following conclusions can be derived from
this study:
1 Localization of an impacted canine with
vertical parallax was more successful than with
magnification, although the difference was not
significant (P > 0.05).
2 There was a wide variation in agreement
between true and predicted position from both
parallax and magnification between different
examiners and dental specialities, not neces-
sarily related to familiarity with the method.
3 Seventy-six per cent of the impacted canines
could be successfully located with vertical
parallax and 66 per cent with magnification.
According to Kappa statistic the prediction
with parallax can only be classified ‘moderate’
and with magnification ‘fair’.
4 Almost 90 per cent of the palatally impacted
canines could be correctly detected with
parallax and magnification, but only 46 per cent
of the buccal canines with parallax and 11 per
cent with magnification. Therefore, magnifi-
cation is only suitable for the detection of
palatally impacted canines. If the canine is
placed buccally and cannot be palpated, an
upper anterior occlusal view is necessary. Since
both techniques are unsatisfactory in the
localization of a buccal canine, additional views
may be justified when a buccally impacted
canine is suspected and cannot be palpated.
Address for correspondence
Miss C. Mason
Eastman Dental Institute for Oral Health Care
Sciences
University of London
256 Gray’s Inn Road
London WC1X 8LD, UK
Acknowledgements
We would like to thank all the consultants,
the senior house officer and the MSc student
who carried out the radiographic examinations.
Without their help this study could not have
been completed. We would also like to thank
the staff in the Day Stay Unit, the Departments
of Radiography and Medical Records for their
co-operation.
References
Broadway R T, Gould D G 1960 Surgical requirements of
the orthodontist. British Dental Journal 108: 1187–1190
RADIOGRAPHIC LOCALIZATION OF CANINES 33
10. Clark C 1909 A method of ascertaining the position of
unerupted teeth by means of film radiographs. Proceedings
of the Royal Society of Medicine 3: 87–90
Coupland M A 1987 Localization of misplaced maxillary
canines: orthopantomograph and PA skull views compared.
American Journal of Orthodontics and Dentofacial
Orthopedics 91: 483–492
Dachi S F, Howell F V 1961 A survey of 3,874 routine
full-mouth radiographs. II. A study of impacted teeth.
Oral Surgery, Oral Medicine, Oral Pathology 14: 1165–1169
Ericson S, Kurol J 1987 Radiographic examination of
ectopically erupting maxillary canines. American Journal
of Orthodontics and Dentofacial Orthopedics 91: 483–492
Ericson S, Kurol J 1988 CT diagnosis of ectopically erupting
maxillary canines—a case report. European Journal of
Orthodontics 10: 115–120
Fox N A, Fletcher G A, Horner K 1995 Localizing
maxillary canines using dental panoramic tomography.
British Dental Journal 179: 416–420
Goaz B W, White S C 1982 Oral radiology—principles and
interpretation, 2nd edn. Mosby Co., St Louis
Hitchin A D 1951 The impacted maxillary canine. Dental
Practitioner 2: 100–103
Jacobs S G 1994 Localization of the unerupted maxillary
canine: additional observations. Australian Orthodontic
Journal 13: 71–75
Kendall G M, Darby S C, Rae S 1981 The level of dental
radiography in Great Britain. British Dental Journal
151: 377–379
Kettle M A 1958 Treatment of the unerupted maxillary
canine. Dental Practitioner and Dental Record 8:
245–255
Keur J J 1986 Radiographic localization techniques.
Australian Dental Journal 31: 86–90
McKay C 1978 The unerupted maxillary canine—an
assessment of the role of surgery in 2500 treated cases.
British Dental Journal 145: 207–210
Ostrofsky M K 1976 Localization of impacted canines with
status-x radiography. Oral Surgery, Oral Medicine, Oral
Pathology 42: 529–533
Rayne J 1969 The unerupted maxillary canine. Dental
Practitioner 19: 194–204
Southall P J, Gravely J F 1987 Radiographic localization
of unerupted teeth in the anterior part of the maxilla:
a survey of methods currently employed. British Journal
of Orthodontics 14: 235–242
Southall P J, Gravely J F 1989 Vertical parallax radiology to
localize an object in the anterior part of the maxilla.
British Journal of Orthodontics 16: 79–83
Thilander B, Jakobsson S O 1968 Local factors in impaction
of maxillary canines. Acta Odontologica Scandinavica
26: 145–168
Turk M H, Katzenell J 1970 Panoramic localization. Oral
Surgery, Oral Medicine, Oral Pathology 29: 212–215
Wolf J E, Mattila K 1979 Localization of impacted maxillary
canines by panoramic tomography. Dentomaxillofacial
Radiology 8: 85–91
34 C. MASON ET AL.